BEA Systems is the middleware magnet of the software industry. Since its inception in January 95, BEA has drawn the majority of middleware systems into its fold, starting with transaction processing monitors and then progressing to an object request broker, messaging software, and finally application server technology. Its appetite for middleware, the so-called software glue that ties together disparate applications and monitors transactions, has been unsurpassed by anyone in its sector. In just under three years, BEA has managed to persuade Novell to part with Tuxedo, bought Top-End from NCR, picked up MessageQ and Object Broker in Digital’s fire sale and, most recently, secured it future in the application server space through the $192m acquisition of San Francisco-based internet start-up, WebLogic.
BEA’s acquisition strategy is fairly unique for a software company but it’s one that has been remarkably effective. Its board has established a $289m software empire in little more than four years on the premise that when it comes to a build versus buy business model, the buy model always wins out. Its predilection towards acquiring key middleware products has enabled it to clear the field of any serious middleware threat. And it now occupies a comfortable lead with a 55% market share with combined sales of Tuxedo and Top End, when closest rival IBM trails way behind at 17% with Encina and CICS 6000, according to The Standish Group.
But investors are still edgy. BEA is by no means insulated from competitive pressures. Furthermore, echoes of last third quarters’ profit warning which sliced its stock in half, still reverberate around Wall Street. BEA sought to offer reassurance to investors and analysts last week, when it reported fourth quarter figures above market expectation.
Earnings per share were up to $0.05 from $0.01 in the year-ago period and revenues increased 53% to $82m (including acquisition write-offs). Furthermore, the sales pipeline problems that marred the third quarter 1999 and have long been of concern to analysts seems to have evaporated. BEA’s CFO, Ed Scott, told analysts that the company had $34m of deferred revenues which could be realized in the first quarter of 2000, thus assuaging fears that it would start the new quarter with a hollow pipe line.
The reassurance worked but only partly. The investment banking community immediately rushed back to their desks to adjust their figures. Goldman Sachs immediately upgraded BEA stock from market perform to market outperform, Lehman Brothers increase its 1999 earnings per share outlook from $20 to $25 and Credit Suisse First Boston revised its 12-month price target to $25. But BEA’s share price failed to return to pre-third quarter levels, when the warning destroyed its share price. Company stock crept up just under two points to $17.25 on the day after the fourth- quarter figures were released.
BEA will sustain its current growth for at least two years. The whole market is growing at a decent pace. I don’t see why the stock price is so low, says Karen Boucher at The Standish Group. (The company’s organic grow rate is 40%.) Perhaps the market is waiting until BEA can deliver a set of financials that are not littered with acquisition charges because if acquisition charges are added back into the company’s full year figures, BEA’s losses actually increased to $51.2m from $22.9m in 1998, on revenues up 73.7% at $289m.
The company also has some near term integration challenges that may blight future financial prospects. When BEA bought MessageQ and Object Broker from Digital, it bought the messaging and object request broker to head off the competition in what was then a rapid growth market for ORBs. The plan was to create a platform combining a transaction processor with an object request broker and messaging broker in the shape of Tuxedo, MessageQ and Object Broker called Iceberg, and subsequently name M3. But when M3 was launched mid-last year having been in the development labs for some two and half years, the market was slow to adopt the high-end system. BEA is only just announcing its first three M3 deployments with Amazon.com and DBG Hanover leading the way with M3.
Furthermore, BEA has still to deliver on its promise to integrate MessageQ with Web Logic Enterprise Edition, the former M3 product which BEA’s re-branded last week. But Web Logic Enterprise Edition is more than a new name. The company plans to marry M3 with the WebLogic application server to give the high-end transaction processing monitor EJB (Enterprise Java Bean) support and Java APIs (application programming interfaces). Web Logic Enterprise Edition is due in the fall. In the next phase of development, BEA will integrate TopEnd with WebLogic Enterprise Edition to provide users with multi-threading, additional industry standard security plug-ins and then add messaging features in the form of an API to MessageQ further out.
These integration projects are not insignificant and there is some doubt whether BEA will be able to rise to the challenge. BEA is not a company that knows how to develop software from scratch and it hasn’t shown much history of product development either. There’s general concern that the company will merely incorporate products under one name but add none of its own magic to ensure they are fully integrated, says Boucher.
From the time being, BEA is relying on its current product portfolio to generate revenue growth. Tuxedo sales still predominate, with WebLogic and M3 accounting for 22% of total revenues in the fourth quarter, according to Steve Brown, the company’s CFO. BEA now has 25,000 Tuxedo customers on maintenance contracts and a customer base that has grown to encompass 3,400 companies. It also added 230 WebLogic customers in the quarter and is making a concerted effort to cross-sell WebLogic Enterprise Edition (formerly M3) and Tuxedo into the WebLogic customer base to provide customers with a back-office system and vice versa. In fact, Tuxedo’s success amongst fiscal institutions has earned it the name Taxedo.
BEA has also unveiled a marketing strategy known as EAI (Enterprise Application Integration) which is designed to boost middleware sales by branding BEA products under the fashionable, and it hopes lucrative, banner of ‘application integration servers’ rather than the distinctly out-moded transaction processing monitor/object request broker label. While the initiative will undoubtedly boost BEA’s visibility in the enterprise software market, the announcement is more of a marketing move than a firmly held technology strategy. Elink, BEA’s EAI product portfolio, is simply its Tuxedo OLTP middleware with application programming interfaces to EAI products from TSI Software and InConcert. ‘Only 20% of Tuxedo or M3 customers use the transaction processors for two phase commit, they are both usually used for application integration anyway, says Boucher at The Standish Group.
That said, it is a clever strategy given the amount of attention application integration has attracted and is likely to be one that provides some room for growth. However, that opportunity won’t exist forever. As the middleware market continues to consolidate there is some suggestion that BEA will be crowded out by the enterprise software giants as the provision of middleware per se ceases to be a viable market. Within the next two years, BEA will face a new threat from Microsoft, Sun and Oracle who are moving towards offering a middleware layer on top of their respective operating systems. Microsoft will ship Transaction Server with Windows 2000, Oracle has included middleware functionality in Oracle 8i and Sun is believed to be incorporating middleware into Solaris. From a customers’ perspective it’s compelling to think that you don’t need to choose and buy separate middleware and that its part of the operating system, says Boucher. It remains to be seen whether BEA is geared up to meet this challenge.